Abstract / Summary
Background: Cardiotoxicity remains a major limitation of contemporary cancer therapies, affecting morbidity and long-term outcomes among cancer survivors. Early identification of patients at increased cardiovascular risk is essential to guide surveillance and preventive strategies. Methods and Results: We performed a narrative review of currently available risk assessment and prediction approaches for cancer therapy-related cardiotoxicity, including consensus-based frameworks, statistically derived prediction models, and emerging machine learning-based approaches. Widely used tools, such as the HFA-ICOS (Heart Failure Association–International Cardio-Oncology Society) risk assessment framework, provide a practical and standardized approach to baseline risk stratification and are increasingly implemented in clinical practice. However, their discriminative performance appears variable across studies, and their applicability across diverse oncologic populations may be limited. Key challenges include heterogeneity in cardiotoxicity definitions, limited external validation, and incomplete integration of dynamic parameters. Overall, models based primarily on clinical variables demonstrate only moderate predictive accuracy, particularly in patients receiving anthracyclines. Conclusions: Current risk stratification strategies provide a useful foundation for clinical decision-making but remain limited in their ability to fully capture individual risk. Future approaches should integrate clinical variables with circulating biomarkers, advanced imaging, and AI to enable more dynamic and individualized risk prediction in cardio-oncology.